Respan Raises $5M in Seed Funding to Expand Proactive AI Observability Platform

March 18th reads like a quiet headline until you realize what just slipped through the market’s fingers. Respan, formerly Keywords AI, just pulled in $5M in seed capital, backed by Gradient Ventures, Y Combinator, Hat-Trick Capital, XIAOXIAO FUND, Antigravity Capital, and Alpen Capital. Andy Li, Raymond Huang, and Hendrix Liu are not chasing noise. They are tuning signal. In a market obsessed with building AI agents, Respan is asking a sharper question: who is watching the watchers when those agents break in production?

Andy Li, Founder and CEO, and Raymond Huang, Founder and CTO, are engineering a control plane that does not just observe behavior, it corrects it. Hendrix Liu, Co-Founder, sits right in the current of that evolution, where observability is no longer a dashboard but a decision engine. Call it Respan because the span is the story. From trace to eval to action, they are stretching the lifecycle of AI systems into something continuous, something accountable. Not reactive logs. Proactive loops. The kind that catch drift before it costs you customers.

The numbers are not decoration, they are direction. More than 100 AI startups and enterprise teams already trust the platform. Over 6.5 million end users are downstream of those systems. More than 1 billion logs processed monthly. Over 2 trillion tokens flowing through the pipe. That is not usage, that is gravity. And when gravity shows up, infrastructure follows. 8x revenue growth year over year in 2025 tells you this is not a science project, it is a system finding product market fit at scale.

What Respan is really building is not just observability, it is accountability for machines that are learning faster than teams can monitor them. The platform connects observability, automated and human evaluations, and an adaptive gateway that turns insight into action. Prompts get updated. Regressions get flagged. Quality does not drift quietly into the night. It gets pulled back into line.

Gradient and Y Combinator did not just fund a company. They funded a layer. The layer that sits between ambition and reality in AI. Because every agent deployed without visibility is a liability waiting to invoice you later. Respan is stepping into that gap with a simple thesis: if AI is going to run the workflow, someone has to run the AI.